Transport & Safety - Enhancing Pedestrian Experiences, Road Crossing Safety, and Accident Prevention in Melbourne

Authored by: Kruthi Shetty

Duration: 120 mins
Level: Intermediate
Pre-requisite Skills: Python, Data Engineering and Analysis

Scenario

As an elderly person my main priority is to move through Melbourne's streets safely, especially while crossing roadways. To ensure my wellbeing and confidence in travelling around, I hope the city would take my particular requirements and worries into consideration.
Personal safety is a pedestrian's main worry when navigating the city, especially when they have to cross roadways. Elderly pedestrians in particular want the city to implement modifications and enhancements that would enhance their general sense of wellbeing and confidence while they move. They want an urban environment that takes their safety concerns and needs into consideration, especially when it comes to crossing roadways. They want to be able to walk about the city with confidence and a sense of security.

As a council ensuring the safety of our residents is a key concern, we are dedicated to investing in traffic safety programmes that have a real and noticeable impact on lowering collisions, injuries, and deaths. We must constantly evaluate how well our present methods for designing the road network are working and see if they are in line with our safety objectives.
The council is aware that seniors face specific demands and difficulties while traversing city streets, notably at junctions and when crossing streets. Reduced mobility, slower walking rates, and possible sensory impairments are only a few of these difficulties. The council is aware that it is crucial to address road safety from the standpoint of an elderly person. The creation of an inclusive and accessible urban environment is under the purview of the council. The council can plan and implement infrastructure upgrades that increase safety and accessibility by taking into account the worries and needs of older pedestrians.

What Will This Use Case Teach You?

At the end of this use case you will:

• Learn how to import the dataset into a Pandas dataframe from different sources.
• Understand how to clean the data by eliminating any pointless columns and dealing with missing values, and organise data in the format needed for visualization.
• Learn how to visualize by utilizing libraries like Matplotlib, Seaborn, or Folium (for interactive maps) to create heatmaps.
• Use the data to create heatmaps to plot the density of pedestrian accidents, traffic intersections, and other relevant data points on the map to visualize high-risk areas and accident hotspots.

Introduction

Melbourne's streets and landmarks are a draw for both visitors and locals since they are located in the centre of a thriving urban dynamic and cultural variety. However, in the middle of all this urban activity, pedestrian safety must come first. In order to improve pedestrian experiences and address the urgent problem of road safety, this use case combines data science with urban design. All pedestrians, particularly the elderly, should feel safe and confident when navigating the streets, especially when crossing roads. Keeping efficient traffic flow and pedestrian safety in balance is a difficult task that calls for a multifaceted strategy.

In this use case, we set out on a data-driven quest to understand pedestrian incidents, pinpoint accident hotspots, and provide suggestions for practical road safety improvements. This study shows how pedestrian age has a substantial impact on accident rates, leading us to consider creative solutions like the Singaporean practise. The idea of giving the elderly and disabled the ability to touch a card on the traffic signal to extend the crossing time has opportunities for improving pedestrian safety.

Pedestrians over 65 years of age are involved in a higher number of crashes because they:

Have more difficulty judging the speed and distance of traffic accurately, as eyesight and hearing begin to decline. May not move or react quickly to avoid approaching vehicles.

Datasets List

• Pedestrian Counting System (counts per hour)
• VicRoads Accident data
• Traffic Signal Data
• Traffic Volume Data

Contents

1. Import important Libraries
2. Connect, Test and Analysis of Datasets
   • 2.1. Pedestrian Counting System (counts per hour)
   • 2.2. VicRoads Accident data
   • 2.3. VicRoads Traffic Signal Data
   • 2.4. VicRoads Traffic Volume Data
3. Combining all Visualisations and Data
4. Findings & Opportunities
5. Conclusions
6. Thank You!

1. Import important Libraries

# Standard
import os
import json

# Data import
import requests

# Data manipulation
import pandas as pd


# Plotting 
import plotly.graph_objs as go
import plotly.express as px
import folium
import folium.plugins as plugins

/Users/kruthishetty/opt/anaconda3/lib/python3.9/site-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected version 1.24.3
  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"

2. Connect, Test and Analysis of Datasets

We must utilise the v2.1 of their API in order to connect to the Melbourne Open Data Portal and collect data. In this way, a custom URL is made using the distinct dataset id (often the name that appears below after /datasets/).

#Pedestrian Counting System (counts per hour)

Ped_count_url= 'https://data.melbourne.vic.gov.au/api/explore/v2.1/catalog/datasets/pedestrian-counting-system-monthly-counts-per-hour/exports/json?limit=-1&offset=0&timezone=UTC'
r = requests.get(Ped_count_url)
response = r.json()
Ped_count = pd.DataFrame(response)

# Print details of data
print(f'The shape of the dataset is:{Ped_count.shape}')
print()
print('The first ten rows of this dataset are:')

# Print the first 10 rows of data
Ped_count.head(10)

The shape of the dataset is:(281567, 5)

The first ten rows of this dataset are:

	timestamp	locationid	direction_1	direction_2	total_of_directions
0	2023-08-30T16:00:00+00:00	1	12	18	30
1	2023-08-30T22:00:00+00:00	1	288	238	526
2	2023-08-30T23:00:00+00:00	1	313	372	685
3	2023-08-31T02:00:00+00:00	1	1270	1001	2271
4	2023-08-31T09:00:00+00:00	1	630	414	1044
5	2023-08-31T10:00:00+00:00	1	433	315	748
6	2023-08-31T11:00:00+00:00	1	310	207	517
7	2023-08-31T13:00:00+00:00	1	54	64	118
8	2023-08-31T14:00:00+00:00	1	38	36	74
9	2023-08-30T14:00:00+00:00	2	4	17	21

As we can see, each record is described by 5 fields The following fields can be divided up for each record:

timestamp: The date and time of the pedestrian count are shown in this column. It will be essential for examining pedestrian activity trends over time and maybe linking accident data.

locationid: This column specifies the precise place where the pedestrian count was made. This makes it easier to identify pedestrian hotspots and provide suggestions for safety improvements since it enables you to link pedestrian counts with various city locations.

direction_1: This column shows the number of people walking in the first direction, such as along one side of the street or across a crosswalk.

direction_2: This column shows the number of people walking in the other direction, such as across the street or across a crosswalk.

total_of_directions: This column shows the aggregate total of pedestrians travelling in both directions. This can be used as a general indicator of foot traffic in the area.

In order to provide insightful data and visuals regarding each dataset, we will now study each one separately. This will help us later on when creating interactive maps and forecasts.

2.1 Pedestrian Counting System (counts per hour)

By presenting the first five rows of the dataset, let's first review it.

Let's first review the dataset by displaying the first five rows of the dataset.

Ped_count.head(5)

	timestamp	locationid	direction_1	direction_2	total_of_directions
0	2023-08-30T16:00:00+00:00	1	12	18	30
1	2023-08-30T22:00:00+00:00	1	288	238	526
2	2023-08-30T23:00:00+00:00	1	313	372	685
3	2023-08-31T02:00:00+00:00	1	1270	1001	2271
4	2023-08-31T09:00:00+00:00	1	630	414	1044

# Convert timestamp to datetime format
Ped_count['timestamp'] = pd.to_datetime(Ped_count['timestamp'])

# Extract day of the week and time of day
Ped_count['day_of_week'] = Ped_count['timestamp'].dt.dayofweek
Ped_count['hour'] = Ped_count['timestamp'].dt.hour

Let us rename the total_direction column for better user experience.

# Rename the 'total_of_directions' column to 'total_pedestrian_count'
Ped_count.rename(columns={'total_of_directions': 'total_pedestrian_count'}, inplace=True)

Let us drop the redundant columns

# Drop 'direction_1' and 'direction_2' columns from the DataFrame
Ped_count.drop(columns=['direction_1', 'direction_2'], inplace=True)

Let's Visualize Number of Features by Pedestrian count.

Ped_count

	timestamp	locationid	total_pedestrian_count	day_of_week	hour
0	2023-08-30 16:00:00+00:00	1	30	2	16
1	2023-08-30 22:00:00+00:00	1	526	2	22
2	2023-08-30 23:00:00+00:00	1	685	2	23
3	2023-08-31 02:00:00+00:00	1	2271	3	2
4	2023-08-31 09:00:00+00:00	1	1044	3	9
...	...	...	...	...	...
281562	2023-09-23 07:00:00+00:00	123	564	5	7
281563	2023-09-23 08:00:00+00:00	123	38	5	8
281564	2023-09-23 09:00:00+00:00	123	3	5	9
281565	2023-09-23 11:00:00+00:00	123	1	5	11
281566	2023-09-23 05:00:00+00:00	124	12	5	5

281567 rows × 5 columns

# Check for missing values in the Ped_count DataFrame
missing_values = Ped_count.isnull().sum()

# Display the count of missing values for each column
print(missing_values)

timestamp                 0
locationid                0
total_pedestrian_count    0
day_of_week               0
hour                      0
dtype: int64

import matplotlib.pyplot as plt
# Group the data by day of the week and calculate the sum of total_pedestrian_count
pedestrian_counts_by_day = Ped_count.groupby('day_of_week')['total_pedestrian_count'].sum()

# Define the names for days of the week (0 = Monday, 6 = Sunday)
day_names = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']

# Create a bar plot
plt.figure(figsize=(10, 6))
plt.bar(day_names, pedestrian_counts_by_day, color='darkblue')
plt.xlabel('Day of the Week')
plt.ylabel('Total Pedestrian Count')
plt.title('Total Pedestrian Count by Day of the Week')
plt.xticks(rotation=45)
plt.tight_layout()

# Show the plot
plt.show()

2.2. VicRoads Accident data- crash stats

import requests
from io import BytesIO
from zipfile import ZipFile

url = 'https://vicroadsopendatastorehouse.vicroads.vic.gov.au/opendata/Road_Safety/ACCIDENT.zip'
content = requests.get(url)

# Checking if the request was successful
if content.status_code == 200:
    # Creating a list of CSV files to extract
    csv_files_to_extract = ['ACCIDENT.csv', 'PERSON.csv', 'NODE.csv', 'ACCIDENT_EVENT.csv', 'ACCIDENT_LOCATION.csv', 'SUBDCA.csv', 'ROAD_SURFACE_COND.csv']

    # Unzip the content
    with ZipFile(BytesIO(content.content)) as zip_file:
        # Loop through the files in the ZIP archive
        for file_name in zip_file.namelist():
            # Check if the file is one of the CSV files you want to extract
            if file_name in csv_files_to_extract:
                # Extract the file to the current directory
                zip_file.extract(file_name)

    print('Selected CSV files extracted successfully.')

# If the request was not successful, handle the error accordingly
else:
    print('Failed to retrieve the ZIP file.')

Selected CSV files extracted successfully.

import pandas as pd
import os

# Define columns to drop from each CSV file
columns_to_drop = {
    'ACCIDENT.csv': ['Day Week Description', 'ACCIDENT_TYPE', 'DIRECTORY', 'EDITION', 'LIGHT_CONDITION', 'NODE_ID', 'NO_PERSONS_KILLED', 'NO_PERSONS_NOT_INJ', 'POLICE_ATTEND', 'ROAD_GEOMETRY', 'Road Geometry Desc', 'SEVERITY', 'SPEED_ZONE' ],
    'PERSON.csv': ['ACCIDENT_NO', 'VEHICLE_ID', 'INJ_LEVEL', 'SEATING_POSITION', 'HELMET_BELT_WORN', 'ROAD_USER_TYPE', 'LICENCE_STATE', 'EJECTED_CODE'],
    'NODE.csv': ['ACCIDENT_NO', 'NODE_ID', 'NODE_TYPE', 'VICGRID94_X', 'VICGRID94_Y', 'LGA_NAME_ALL', 'REGION_NAME'],
    'ACCIDENT_EVENT.csv': ['ACCIDENT_NO', 'EVENT_SEQ_NO', 'EVENT_TYPE', 'VEHICLE_1_ID', 'VEHICLE_1_COLL_PT', 'Vehicle 1 Coll Pt Desc', 'VEHICLE_2_ID', 'VEHICLE_2_COLL_PT', 'Vehicle 2 Coll Pt Desc', 'PERSON_ID', 'OBJECT_TYPE'],
    'ACCIDENT_LOCATION.csv': ['ACCIDENT_NO', 'NODE_ID', 'ROAD_ROUTE_1', 'ROAD_TYPE_INT', 'DISTANCE_LOCATION', 'DIRECTION_LOCATION', 'NEAREST_KM_POST', 'OFF_ROAD_LOCATION'],
    'SUBDCA.csv' : ['Sub Dca Code Desc'],
    'ROAD_SURFACE_COND.csv' : ['Surface Cond Desc']
    # Add more CSV files and columns to drop as needed
}

# Initialize an empty list to store DataFrames
data_frames = []

# Loop through each CSV file and get information
for file_name in csv_files_to_extract:
    try:
        # Check if this file has columns to drop
        if file_name in columns_to_drop:
            # Load the CSV file into a Pandas DataFrame
            df = pd.read_csv(file_name)
            # Drop the specified columns
            df.drop(columns=columns_to_drop[file_name], inplace=True)
        else:
            # Load the CSV file without dropping columns
            df = pd.read_csv(file_name)
        
        # Append the DataFrame to the list
        data_frames.append(df)
    except FileNotFoundError:
        print(f"File not found: {file_name}")
    except Exception as e:
        print(f"Error processing {file_name}: {e}")

# Concatenate all DataFrames into a single DataFrame

accident_pedestrians = pd.concat(data_frames, axis =1)
accident_pedestrians.info()

/var/folders/nm/ntbwszps2j377flx0d25z84w0000gn/T/ipykernel_46276/2886532505.py:25: DtypeWarning: Columns (10,13) have mixed types. Specify dtype option on import or set low_memory=False.
  df = pd.read_csv(file_name)
/var/folders/nm/ntbwszps2j377flx0d25z84w0000gn/T/ipykernel_46276/2886532505.py:25: DtypeWarning: Columns (6,9,10,13) have mixed types. Specify dtype option on import or set low_memory=False.
  df = pd.read_csv(file_name)

<class 'pandas.core.frame.DataFrame'>
Int64Index: 490948 entries, 0 to 490947
Data columns (total 40 columns):
 #   Column                Non-Null Count   Dtype  
---  ------                --------------   -----  
 0   ACCIDENT_NO           203708 non-null  object 
 1   ACCIDENTDATE          203708 non-null  object 
 2   ACCIDENTTIME          203708 non-null  object 
 3   Accident Type Desc    203708 non-null  object 
 4   DAY_OF_WEEK           203708 non-null  float64
 5   DCA_CODE              203708 non-null  float64
 6   DCA Description       203708 non-null  object 
 7   PAGE                  201360 non-null  object 
 8   GRID_REFERENCE_X      201360 non-null  object 
 9   GRID_REFERENCE_Y      201360 non-null  object 
 10  Light Condition Desc  203708 non-null  object 
 11  NO_OF_VEHICLES        203708 non-null  float64
 12  NO_PERSONS            203708 non-null  float64
 13  NO_PERSONS_INJ_2      203708 non-null  float64
 14  NO_PERSONS_INJ_3      203708 non-null  float64
 15  PERSON_ID             490948 non-null  object 
 16  SEX                   490948 non-null  object 
 17  AGE                   469319 non-null  float64
 18  Age Group             490948 non-null  object 
 19  Inj Level Desc        490948 non-null  object 
 20  Road User Type Desc   490948 non-null  object 
 21  PEDEST_MOVEMENT       490948 non-null  object 
 22  POSTCODE              416854 non-null  float64
 23  TAKEN_HOSPITAL        490948 non-null  object 
 24  LGA_NAME              221797 non-null  object 
 25  DEG_URBAN_NAME        221797 non-null  object 
 26  Lat                   221797 non-null  float64
 27  Long                  221797 non-null  float64
 28  POSTCODE_NO           221797 non-null  float64
 29  Event Type Desc       326638 non-null  object 
 30  Object Type Desc      326638 non-null  object 
 31  ROAD_NAME             202702 non-null  object 
 32  ROAD_TYPE             200309 non-null  object 
 33  ROAD_NAME_INT         201886 non-null  object 
 34  ACCIDENT_NO           290966 non-null  object 
 35  SUB_DCA_CODE          290966 non-null  object 
 36  SUB_DCA_SEQ           290966 non-null  float64
 37  ACCIDENT_NO           205030 non-null  object 
 38  SURFACE_COND          205030 non-null  float64
 39  SURFACE_COND_SEQ      205030 non-null  float64
dtypes: float64(14), object(26)
memory usage: 153.6+ MB

import pandas as pd
import re
import numpy as np

# Remove extra spaces and irregularities from ACCIDENTTIME using regex
accident_pedestrians['ACCIDENTTIME'] = accident_pedestrians['ACCIDENTTIME'].apply(lambda x: re.sub(r'\s+', ' ', str(x)).strip())

# Combine ACCIDENTDATE and ACCIDENTTIME and parse as datetime
accident_pedestrians['DATETIME'] = pd.to_datetime(accident_pedestrians['ACCIDENTDATE'] + ' ' + accident_pedestrians['ACCIDENTTIME'], format='%d/%m/%Y %H:%M:%S')

# Create new features from the accident date variable
accident_pedestrians['NUMDayOfWeek'] = accident_pedestrians['DATETIME'].dt.strftime('%w')
accident_pedestrians['NUMWeekOfYear'] = accident_pedestrians['DATETIME'].dt.strftime('%W')
accident_pedestrians['NUMDayOfYear'] = accident_pedestrians['DATETIME'].dt.strftime('%j')
accident_pedestrians['NUMYearOfAcc'] = accident_pedestrians['DATETIME'].dt.strftime('%Y')

# Clean up text white space in the DCA Description variable
accident_pedestrians['DCA Description'] = accident_pedestrians['DCA Description'].str.rstrip()

# Create and apply a group mapping for the hour of accident
hour_mapping = {
    '00': 'Early Morning', '01': 'Early Morning', '02': 'Early Morning', '03': 'Early Morning', '04': 'Early Morning',
    '05': 'Early Morning', '06': 'Morning', '07': 'Morning', '08': 'Morning', '09': 'Late Morning', '10': 'Late Morning',
    '11': 'Late Morning', '12': 'Early Afternoon', '13': 'Early Afternoon', '14': 'Early Afternoon', '15': 'Late Afternoon',
    '16': 'Late Afternoon', '17': 'Evening', '18': 'Evening', '19': 'Evening', '20': 'Night', '21': 'Night', '22': 'Night',
    '23': 'Night'
}
accident_pedestrians['hourgroup'] = accident_pedestrians['DATETIME'].dt.strftime('%H').map(hour_mapping)

# Convert all categorical variables to strings

categorical_columns =  ['Accident Type Desc', 'DAY_OF_WEEK', 'DCA Description', 'PAGE', 'GRID_REFERENCE_X', 'GRID_REFERENCE_Y',
                       'Light Condition Desc', 'SEX', 'Age Group', 'Inj Level Desc', 'Road User Type Desc', 'PEDEST_MOVEMENT',
                       'TAKEN_HOSPITAL', 'LGA_NAME', 'DEG_URBAN_NAME','POSTCODE', 'POSTCODE_NO',  'Event Type Desc', 'Object Type Desc', 'ROAD_NAME',
                       'ROAD_TYPE', 'ROAD_NAME_INT', 'SUB_DCA_CODE', 'SURFACE_COND']

accident_pedestrians[categorical_columns] = accident_pedestrians[categorical_columns].astype(str)


# Replace non-integer values with NaN in numerical columns
numerical_columns = ['DCA_CODE', 'NO_OF_VEHICLES', 'NO_PERSONS', 'NO_PERSONS_INJ_2', 'NO_PERSONS_INJ_3',
                     'AGE', 'SUB_DCA_SEQ', 'SURFACE_COND', 'SURFACE_COND_SEQ']
accident_pedestrians[numerical_columns] = accident_pedestrians[numerical_columns].apply(pd.to_numeric, errors='coerce')

# Convert numerical columns to integers, handling missing values
accident_pedestrians[numerical_columns] = accident_pedestrians[numerical_columns].fillna(0).astype(int)

# Create a new feature by concatenating ROAD_NAME and ROAD_TYPE
accident_pedestrians['CATAccidentRoadGroup'] = accident_pedestrians['ROAD_NAME'].fillna('') + ' ' + accident_pedestrians['ROAD_TYPE'].fillna('')

# Print the information summary for the cleaned dataset
print(accident_pedestrians.info())

<class 'pandas.core.frame.DataFrame'>
Int64Index: 490948 entries, 0 to 490947
Data columns (total 47 columns):
 #   Column                Non-Null Count   Dtype         
---  ------                --------------   -----         
 0   ACCIDENT_NO           203708 non-null  object        
 1   ACCIDENTDATE          203708 non-null  object        
 2   ACCIDENTTIME          490948 non-null  object        
 3   Accident Type Desc    490948 non-null  object        
 4   DAY_OF_WEEK           490948 non-null  object        
 5   DCA_CODE              490948 non-null  int64         
 6   DCA Description       490948 non-null  object        
 7   PAGE                  490948 non-null  object        
 8   GRID_REFERENCE_X      490948 non-null  object        
 9   GRID_REFERENCE_Y      490948 non-null  object        
 10  Light Condition Desc  490948 non-null  object        
 11  NO_OF_VEHICLES        490948 non-null  int64         
 12  NO_PERSONS            490948 non-null  int64         
 13  NO_PERSONS_INJ_2      490948 non-null  int64         
 14  NO_PERSONS_INJ_3      490948 non-null  int64         
 15  PERSON_ID             490948 non-null  object        
 16  SEX                   490948 non-null  object        
 17  AGE                   490948 non-null  int64         
 18  Age Group             490948 non-null  object        
 19  Inj Level Desc        490948 non-null  object        
 20  Road User Type Desc   490948 non-null  object        
 21  PEDEST_MOVEMENT       490948 non-null  object        
 22  POSTCODE              490948 non-null  object        
 23  TAKEN_HOSPITAL        490948 non-null  object        
 24  LGA_NAME              490948 non-null  object        
 25  DEG_URBAN_NAME        490948 non-null  object        
 26  Lat                   221797 non-null  float64       
 27  Long                  221797 non-null  float64       
 28  POSTCODE_NO           490948 non-null  object        
 29  Event Type Desc       490948 non-null  object        
 30  Object Type Desc      490948 non-null  object        
 31  ROAD_NAME             490948 non-null  object        
 32  ROAD_TYPE             490948 non-null  object        
 33  ROAD_NAME_INT         490948 non-null  object        
 34  ACCIDENT_NO           290966 non-null  object        
 35  SUB_DCA_CODE          490948 non-null  object        
 36  SUB_DCA_SEQ           490948 non-null  int64         
 37  ACCIDENT_NO           205030 non-null  object        
 38  SURFACE_COND          490948 non-null  int64         
 39  SURFACE_COND_SEQ      490948 non-null  int64         
 40  DATETIME              203708 non-null  datetime64[ns]
 41  NUMDayOfWeek          203708 non-null  object        
 42  NUMWeekOfYear         203708 non-null  object        
 43  NUMDayOfYear          203708 non-null  object        
 44  NUMYearOfAcc          203708 non-null  object        
 45  hourgroup             203708 non-null  object        
 46  CATAccidentRoadGroup  490948 non-null  object        
dtypes: datetime64[ns](1), float64(2), int64(9), object(35)
memory usage: 179.8+ MB
None

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

age_groups_order = ['64-69', '70+' ]

# 1. Age Distribution of Pedestrians Involved in Accidents (Histogram)
plt.figure(figsize=(10, 6))
sns.histplot(data=accident_pedestrians, x='AGE', bins=20, kde=True, color='skyblue')
plt.title('Age Distribution of Pedestrians Involved in Accidents')
plt.xlabel('Age')
plt.ylabel('Frequency')
plt.show()

# 2. Accident Severity by Age Group (Bar Chart)
plt.figure(figsize=(10, 6))
sns.countplot(data=accident_pedestrians, x='Age Group', hue='Inj Level Desc', order=age_groups_order,
              palette='Set2')
plt.title('Accident Severity by Age Group')
plt.xlabel('Age Group')
plt.ylabel('Count')
plt.legend(title='Injury Level', loc='upper right')
plt.xticks(rotation=45)
plt.show()

# 3. Accident Types for Pedestrians (Bar Chart)
plt.figure(figsize=(10, 6))
sns.countplot(data=accident_pedestrians, y='Accident Type Desc', palette='viridis')
plt.title('Accident Types for Pedestrians')
plt.xlabel('Count')
plt.ylabel('Accident Type')
plt.show()


# 4. Accident Hotspots (Scatter Plot on Map - Geospatial Plot)
plt.figure(figsize=(12, 10))
sns.scatterplot(data=accident_pedestrians, x='Long', y='Lat', alpha=0.2)
plt.title('Accident Hotspots on Map')
plt.xlabel('Longitude')
plt.ylabel('Latitude')
plt.show()

# Filter the data to include only pedestrians
pedestrian_data = accident_pedestrians[accident_pedestrians['Road User Type Desc'] == 'Pedestrians']

# Filter the pedestrian data to include only pedestrians over the age of 65
pedestrians_over_65 = pedestrian_data[(pedestrian_data['Age Group'] == '64-69') | (pedestrian_data['Age Group'] == '70+')]

# Check if there is data for pedestrians over 65
if not pedestrians_over_65.empty:
    # Create a bar chart to visualize accident types for pedestrians over 65
    plt.figure(figsize=(10, 6))
    sns.countplot(data=pedestrians_over_65, x='Accident Type Desc', palette='viridis')
    plt.title('Accident Types for Pedestrians Over 65')
    plt.xlabel('Count')
    plt.ylabel('Accident Type')
    plt.xticks(rotation=45)
    plt.show()
else:
    print("No data available for pedestrians over the age of 65.")

2.3. Traffic Signal Data

import geopandas as gpd

# URL of the GeoJSON file
url2 = "https://vicroadsopendata-vicroadsmaps.opendata.arcgis.com/datasets/1f3cb954526b471596dbffa30e56bb32_0.geojson?outSR=%7B%22latestWkid%22%3A3111%2C%22wkid%22%3A102171%7D"

# Read the GeoJSON file from the URL into a GeoDataFrame
traffic_lights = gpd.read_file(url2)

# Now, 'gdf' contains your spatial data as a GeoDataFrame
traffic_lights

	OBJECTID	TLIGHTS_	TLIGHTS_ID	SITE_NO	SITE_NAME	SITE_TYPE	DIRECTORY	DIR_REF	D_ADDED	D_TOWNS	D_EDITED	D_REMOVED	LINK_MODE	STATUS	COMMENTS	MULTI	UFI	ARC_UFI	geometry
0	45042	NaN	NaN	5168	BARWON HEADS ROAD/WARALILY BVD	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (144.37788 -38.23666)
1	45043	NaN	NaN	2015	MELBOURNE NR HOBART (KNOX CITY)	POS			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (145.24197 -37.86925)
2	45044	NaN	NaN	6273	MITCHELL/QUEEN	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (144.28034 -36.76131)
3	45045	NaN	NaN	520	MOOROODUC/SAGES	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL	Bluetooth	NaN	NaN	NaN	POINT (145.13371 -38.19567)
4	45046	NaN	NaN	3629	CHURCHILL NR BRAYBROOK COMMUNITY CENTRE	POS			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (144.85338 -37.78613)
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4847	49889	NaN	NaN	5335	PRINCES HIGHWAY WEST/HESSE STREET	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (144.32642 -38.21496)
4848	49890	NaN	NaN	5336	PRINCES HWY WEST/DEANS MARSH RD/MURRELL	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL		NaN	NaN	NaN	POINT (144.32642 -38.21496)
4849	49891	NaN	NaN	2996	MELROSE DRIVE/GRANTS ROAD/CENTRE ROAD	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL	Bluetooth	NaN	NaN	NaN	POINT (144.85216 -37.67554)
4850	49892	NaN	NaN	1559	Mornington Peninsula Fwy/Nepean Hwy	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL	METER,UPS	NaN	NaN	NaN	POINT (144.99103 -38.33009)
4851	49893	NaN	NaN	5339	DUNCANS/K/PRINCES FWY OFF RAMP	INT			NaN	NaN	NaN	NaN	NaN	OPERATIONAL	PTS_SIGN	NaN	NaN	NaN	POINT (144.68333 -37.91302)

4852 rows × 19 columns

# Access latitude and longitude from the 'geometry' column
traffic_lights['latitude'] = traffic_lights['geometry'].y
traffic_lights['longitude'] = traffic_lights['geometry'].x
#Dropping the unwanted columns
columns_to_drop = ['TLIGHTS_', 'TLIGHTS_ID', 'DIRECTORY', 'DIR_REF', 'D_ADDED', 'D_TOWNS', 'D_EDITED', 'D_REMOVED', 'LINK_MODE', 'MULTI', 'UFI', 'ARC_UFI', 'geometry']
traffic_lights.drop(columns=columns_to_drop, inplace= True )

traffic_lights

	OBJECTID	SITE_NO	SITE_NAME	SITE_TYPE	STATUS	COMMENTS	latitude	longitude
0	45042	5168	BARWON HEADS ROAD/WARALILY BVD	INT	OPERATIONAL		-38.236657	144.377875
1	45043	2015	MELBOURNE NR HOBART (KNOX CITY)	POS	OPERATIONAL		-37.869250	145.241974
2	45044	6273	MITCHELL/QUEEN	INT	OPERATIONAL		-36.761312	144.280339
3	45045	520	MOOROODUC/SAGES	INT	OPERATIONAL	Bluetooth	-38.195666	145.133709
4	45046	3629	CHURCHILL NR BRAYBROOK COMMUNITY CENTRE	POS	OPERATIONAL		-37.786134	144.853384
...	...	...	...	...	...	...	...	...
4847	49889	5335	PRINCES HIGHWAY WEST/HESSE STREET	INT	OPERATIONAL		-38.214960	144.326416
4848	49890	5336	PRINCES HWY WEST/DEANS MARSH RD/MURRELL	INT	OPERATIONAL		-38.214960	144.326416
4849	49891	2996	MELROSE DRIVE/GRANTS ROAD/CENTRE ROAD	INT	OPERATIONAL	Bluetooth	-37.675543	144.852164
4850	49892	1559	Mornington Peninsula Fwy/Nepean Hwy	INT	OPERATIONAL	METER,UPS	-38.330091	144.991029
4851	49893	5339	DUNCANS/K/PRINCES FWY OFF RAMP	INT	OPERATIONAL	PTS_SIGN	-37.913023	144.683332

4852 rows × 8 columns

# Group data by SITE_TYPE and count the occurrences
site_type_counts = traffic_lights['SITE_TYPE'].value_counts().reset_index()
site_type_counts.columns = ['SITE_TYPE', 'count']

# Create a pie chart using Plotly Express
fig = px.pie(site_type_counts, names='SITE_TYPE', values='count', title='Distribution of SITE_TYPE')
fig.show()

2.4. Traffic Volume Data

import pandas as pd

# URL of the CSV file
url3 = "https://vicroadsopendata-vicroadsmaps.opendata.arcgis.com/datasets/5512df2ff41e4941bacf868053dbfba9_0.csv?outSR=%7B%22latestWkid%22%3A3111%2C%22wkid%22%3A102171%7D"

# Read the CSV file from the URL
traffic_volume = pd.read_csv(url3)


traffic_volume

	OBJECTID_1	OBJECTID	TIS_ID	HMGNS_FLOW_ID	HMGNS_LNK_ID	HMGNS_LNK_DESC	LGA_SHORT_NM	RGN_LONG_NM	ROAD_NBR	DECLARED_ROAD_NM	...	TWO_WAY_AADT_TRUCKS	ALLVEH_AMPEAK_AADT	ALLVEH_PMPEAK_AADT	GROWTH_RATE	CI	AM_PEAK_SPEED	OFF_PEAK_SPEED	PM_PEAK_SPEED	YR	LABEL
0	1	743	14915	14915	2006	MARYSVILLE-WOODS POINT ROAD btwn LAKE MOUNTAI...	YARRA RANGES	METROPOLITAN SOUTH EAST REGION	4961	MARYSVILLE-WOODS POINT ROAD	...	0.0	NaN	NaN	0.013	0.005	NaN	NaN	NaN	2020	24* (13% 3*) EAST BOUND
1	2	650	14140	14140	8786	STEELS CREEK ROAD btwn WILLOWBEND DRIVE & ELT...	YARRA RANGES	METROPOLITAN SOUTH EAST REGION	9999	Not Applicable	...	40.0	NaN	NaN	0.019	0.002	NaN	NaN	NaN	2020	373* (6% 22*) NORTH BOUND
2	3	701	12113	12113	6035	LATROBE ROAD btwn TANJIL EAST ROAD & GORDON S...	LATROBE	EASTERN REGION	5911	MORWELL-YALLOURN NORTH ROAD	...	160.0	NaN	NaN	0.015	0.009	NaN	NaN	NaN	2020	1,100* (6% 61*) NORTH BOUND
3	4	702	12897	12897	7079	CASTERTON ROAD btwn GLENELG HIGHWAY & COLERAI...	SOUTHERN GRAMPIANS	SOUTH WESTERN REGION	2670	GLENELG HIGHWAY	...	340.0	NaN	NaN	0.020	0.001	NaN	NaN	NaN	2020	801* (21% 165*) WEST BOUND
4	5	703	9893	9893	3475	HUTTON ROAD btwn CHAPEL ROAD & GREENS ROAD	DANDENONG	METROPOLITAN SOUTH EAST REGION	5168	BRAESIDE-DANDENONG ROAD	...	1500.0	1000.0	1100.0	0.003	0.002	NaN	NaN	NaN	2020	12,000 (6% 744*) WEST BOUND
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14669	14670	33800	16426	16426	8864	WENTWORTH STREET BTWN SOUTH GIPPSLAND HIGHWAY ...	CASEY	METROPOLITAN SOUTH EAST REGION	0	Not Applicable	...	NaN	NaN	NaN	0.030	0.003	NaN	NaN	NaN	2020	4,000* (N/A) EAST BOUND
14670	14671	33809	16440	16440	3371	BALLARTO RD BTWN POTTS RD AND MCCORMICKS RD	FRANKSTON	METROPOLITAN SOUTH EAST REGION	5152	BALLARTO ROAD	...	1500.0	950.0	563.0	0.016	0.002	NaN	NaN	NaN	2020	7,900 (8% 656*) WEST BOUND
14671	14672	33405	16401	16401	1367	ALEXANDRA PDE BTWN GOLD ST AND HODDLE ST	YARRA	METROPOLITAN NORTH WEST REGION	8002	2830F 2080B	...	250.0	NaN	NaN	0.001	0.002	NaN	NaN	NaN	2020	2,400* (5% 126*) EAST BOUND
14672	14673	33406	16402	16402	1367	ALEXANDRA PDE BTWN GOLD ST AND HODDLE ST	YARRA	METROPOLITAN NORTH WEST REGION	8002	2830F 2080B	...	250.0	NaN	NaN	0.001	0.002	NaN	NaN	NaN	2020	2,400* (5% 126*) EAST BOUND
14673	14674	33824	16399	16399	8842	HUON PARK RD BTWN SOUTH GIPPSLAND HWY AND HAZL...	CASEY	METROPOLITAN SOUTH EAST REGION	0	Not Applicable	...	NaN	NaN	NaN	0.030	0.003	NaN	NaN	NaN	2020	4,000* (N/A) WEST BOUND

14674 rows × 50 columns

# List of columns to keep
columns_to_keep = ['HMGNS_LNK_DESC','ROAD_NBR', 'DECLARED_ROAD_NM', 'ALLVEH_AMPEAK_AADT', 'ALLVEH_PMPEAK_AADT', 'AM_PEAK_SPEED', 'OFF_PEAK_SPEED', 'PM_PEAK_SPEED','GROWTH_RATE',  'YR', 'LABEL']

# Remove unwanted columnsALLVEH_AMPEAK_AADT
traffic_volume = traffic_volume[columns_to_keep]

traffic_volume

	HMGNS_LNK_DESC	ROAD_NBR	DECLARED_ROAD_NM	ALLVEH_AMPEAK_AADT	ALLVEH_PMPEAK_AADT	AM_PEAK_SPEED	OFF_PEAK_SPEED	PM_PEAK_SPEED	GROWTH_RATE	YR	LABEL
0	MARYSVILLE-WOODS POINT ROAD btwn LAKE MOUNTAI...	4961	MARYSVILLE-WOODS POINT ROAD	NaN	NaN	NaN	NaN	NaN	0.013	2020	24* (13% 3*) EAST BOUND
1	STEELS CREEK ROAD btwn WILLOWBEND DRIVE & ELT...	9999	Not Applicable	NaN	NaN	NaN	NaN	NaN	0.019	2020	373* (6% 22*) NORTH BOUND
2	LATROBE ROAD btwn TANJIL EAST ROAD & GORDON S...	5911	MORWELL-YALLOURN NORTH ROAD	NaN	NaN	NaN	NaN	NaN	0.015	2020	1,100* (6% 61*) NORTH BOUND
3	CASTERTON ROAD btwn GLENELG HIGHWAY & COLERAI...	2670	GLENELG HIGHWAY	NaN	NaN	NaN	NaN	NaN	0.020	2020	801* (21% 165*) WEST BOUND
4	HUTTON ROAD btwn CHAPEL ROAD & GREENS ROAD	5168	BRAESIDE-DANDENONG ROAD	1000.0	1100.0	NaN	NaN	NaN	0.003	2020	12,000 (6% 744*) WEST BOUND
...	...	...	...	...	...	...	...	...	...	...	...
14669	WENTWORTH STREET BTWN SOUTH GIPPSLAND HIGHWAY ...	0	Not Applicable	NaN	NaN	NaN	NaN	NaN	0.030	2020	4,000* (N/A) EAST BOUND
14670	BALLARTO RD BTWN POTTS RD AND MCCORMICKS RD	5152	BALLARTO ROAD	950.0	563.0	NaN	NaN	NaN	0.016	2020	7,900 (8% 656*) WEST BOUND
14671	ALEXANDRA PDE BTWN GOLD ST AND HODDLE ST	8002	2830F 2080B	NaN	NaN	NaN	NaN	NaN	0.001	2020	2,400* (5% 126*) EAST BOUND
14672	ALEXANDRA PDE BTWN GOLD ST AND HODDLE ST	8002	2830F 2080B	NaN	NaN	NaN	NaN	NaN	0.001	2020	2,400* (5% 126*) EAST BOUND
14673	HUON PARK RD BTWN SOUTH GIPPSLAND HWY AND HAZL...	0	Not Applicable	NaN	NaN	NaN	NaN	NaN	0.030	2020	4,000* (N/A) WEST BOUND

14674 rows × 11 columns

# Rename the "DECLARED_ROAD_NM" column to "Road_Name"
traffic_volume.rename(columns={'DECLARED_ROAD_NM': 'ROAD_NAME'}, inplace=True)
traffic_volume.rename(columns={'YR': 'Year'}, inplace=True)

/var/folders/nm/ntbwszps2j377flx0d25z84w0000gn/T/ipykernel_46276/1108220500.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

/var/folders/nm/ntbwszps2j377flx0d25z84w0000gn/T/ipykernel_46276/1108220500.py:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

Average Growth of Traffic on different roads

# Group the data by Road_Name and calculate the average growth rate
road_growth_data = traffic_volume.groupby('ROAD_NAME')['GROWTH_RATE'].mean()

# Create a line plot
plt.figure(figsize=(10, 6))
road_growth_data.plot(kind='line', marker='o')
plt.title('Average Growth Rate of Traffic on Different Roads Over the Years')
plt.xlabel('Road Name')
plt.ylabel('Average Growth Rate')
plt.xticks(rotation=90)
plt.grid(True)
plt.tight_layout()
plt.show()

After going through pre-processing and feature engineering, the datasets are now prepared for additional analysis.

3. Combining all Visualisations and Data

import pandas as pd

# Filter the data to include only pedestrians
pedestrian_data = accident_pedestrians[accident_pedestrians['Road User Type Desc'] == 'Pedestrians']

# Filter the pedestrian data to include only pedestrians over the age of 65
pedestrians_over_65 = pedestrian_data[(pedestrian_data['Age Group'] == '64-69') | (pedestrian_data['Age Group'] == '70+')]

# Select relevant columns from the pedestrian_data DataFrame
selected_pedestrian_data = pedestrians_over_65[['Lat', 'Long', 'ROAD_NAME', 'NUMYearOfAcc', 'Age Group']]

# Rename columns to match those in the Ped_count dataset
selected_pedestrian_data = selected_pedestrian_data.rename(columns={'Lat': 'latitude', 'Long': 'longitude', 'ROAD_NAME': 'Road Name', 'NUMYearOfAcc': 'Year'})

# Create a list of additional columns from traffic_lights and traffic_volume datasets to add
selected_traffic_lights = ['SITE_NAME', 'STATUS', 'SITE_TYPE']
selected_traffic_volume = ['ALLVEH_AMPEAK_AADT', 'ALLVEH_PMPEAK_AADT', 'GROWTH_RATE']

# Concatenate selected_pedestrian_data and additional columns as separate columns
combined_data = pd.concat([selected_pedestrian_data] + [Ped_count[['total_pedestrian_count']]] + [traffic_lights[selected_traffic_lights]] + [traffic_volume[selected_traffic_volume]], axis=1)

# Include 'IsPedestrian' column and set it to True for pedestrian-related data
combined_data['Pedestrian'] = True

# Display the resulting combined DataFrame
combined_data.head()

	latitude	longitude	Road Name	Year	Age Group	total_pedestrian_count	SITE_NAME	STATUS	SITE_TYPE	ALLVEH_AMPEAK_AADT	ALLVEH_PMPEAK_AADT	GROWTH_RATE	Pedestrian
267	-37.811005	145.237372	BEACH	2006	64-69	400.0	THOMPSON NR VISTULA	OPERATIONAL	POS	NaN	NaN	-0.008	True
452	-38.066280	145.467279	COLAC-FORREST	2006	70+	1849.0	GILBERT/REGENT	OPERATIONAL	INT	1000.0	1100.0	0.019	True
897	-37.848228	144.977528	YORK	2006	70+	3030.0	EASTERN NR GOLD (WEST BOUND)	OPERATIONAL	POS	NaN	NaN	0.019	True
960	-37.875055	144.718164	BELL	2006	70+	137.0	DERRIMUT/BAGGOTT	OPERATIONAL	INT	NaN	NaN	0.026	True
1081	-37.932947	145.034514	MAIN	2006	70+	170.0	HIGH (MCIVOR HWY)/SHORT	OPERATIONAL	INT	NaN	NaN	0.001	True

combined_df_cleaned = combined_data.fillna('N.A')
combined_df_cleaned

	latitude	longitude	Road Name	Year	Age Group	total_pedestrian_count	SITE_NAME	STATUS	SITE_TYPE	ALLVEH_AMPEAK_AADT	ALLVEH_PMPEAK_AADT	GROWTH_RATE	Pedestrian
267	-37.811005	145.237372	BEACH	2006	64-69	400.0	THOMPSON NR VISTULA	OPERATIONAL	POS	N.A	N.A	-0.008	True
452	-38.06628	145.467279	COLAC-FORREST	2006	70+	1849.0	GILBERT/REGENT	OPERATIONAL	INT	1000.0	1100.0	0.019	True
897	-37.848228	144.977528	YORK	2006	70+	3030.0	EASTERN NR GOLD (WEST BOUND)	OPERATIONAL	POS	N.A	N.A	0.019	True
960	-37.875055	144.718164	BELL	2006	70+	137.0	DERRIMUT/BAGGOTT	OPERATIONAL	INT	N.A	N.A	0.026	True
1081	-37.932947	145.034514	MAIN	2006	70+	170.0	HIGH (MCIVOR HWY)/SHORT	OPERATIONAL	INT	N.A	N.A	0.001	True
...	...	...	...	...	...	...	...	...	...	...	...	...	...
281562	N.A	N.A	N.A	N.A	N.A	564.0	N.A	N.A	N.A	N.A	N.A	N.A	True
281563	N.A	N.A	N.A	N.A	N.A	38.0	N.A	N.A	N.A	N.A	N.A	N.A	True
281564	N.A	N.A	N.A	N.A	N.A	3.0	N.A	N.A	N.A	N.A	N.A	N.A	True
281565	N.A	N.A	N.A	N.A	N.A	1.0	N.A	N.A	N.A	N.A	N.A	N.A	True
281566	N.A	N.A	N.A	N.A	N.A	12.0	N.A	N.A	N.A	N.A	N.A	N.A	True

283219 rows × 13 columns

Visualizing a Map with Traffic Signals and Pedestrians over 64-65 and 70+.

import folium
import math

# Filter out rows with missing or invalid latitude and longitude values
combined_df_cleaned = combined_df_cleaned[(combined_df_cleaned['latitude'] != 'N.A') & (combined_df_cleaned['longitude'] != 'N.A')]

# Create a base map
m = folium.Map(location=[-37.811005, 145.237372], zoom_start=20, width='100%', height='100%')

# Define an offset for pedestrian markers to avoid overlap with traffic signals
pedestrian_offset = 0.0001  # You can adjust this value as needed

for index, row in combined_df_cleaned.iterrows():
    lat = float(row['latitude'])
    lon = float(row['longitude'])
    year = row['Year']
    pedestrian_count = row['total_pedestrian_count']
    site_name = row['SITE_NAME']

    # Check if the incident involves pedestrians over 70
    if row['Age Group'] == '70+':
        # Create a pop-up label for incidents involving pedestrians over 70
        popup = f"Year: {year}<br>Location: {row['Road Name']}<br>Pedestrian Count: {pedestrian_count}<br>Age Group: 70+"
        # Use a different icon or marker style for these incidents
        icon_color = 'red'  # Choose a suitable color
        icon_html = f'<i class="fa fa-user" style="color: {icon_color}; font-size: 24px;"></i>'

        # Add the pedestrian marker to the map with an offset
        folium.Marker(
            [lat + pedestrian_offset, lon + pedestrian_offset],
            popup=folium.Popup(popup, max_width=300),
            icon=folium.DivIcon(html=icon_html)
        ).add_to(m)
    elif row['Age Group'] == '64-69':
        # Create a pop-up label for incidents involving pedestrians in the age group "64-69"
        popup = f"Year: {year}<br>Location: {row['Road Name']}<br>Pedestrian Count: {pedestrian_count}<br>Age Group: 64-69"
        # Use blue color for this age group
        icon_color = 'blue'  # Choose a suitable color
        icon_html = f'<i class="fa fa-user" style="color: {icon_color}; font-size: 24px;"></i>'
        
        # Add the pedestrian marker to the map with an offset
        folium.Marker(
            [lat + pedestrian_offset, lon + pedestrian_offset],
            popup=folium.Popup(popup, max_width=300),
            icon=folium.DivIcon(html=icon_html)
        ).add_to(m)
    else:
        # For other age groups, use a default style
        popup = f"Year: {year}<br>Location: {row['Road Name']}<br>Pedestrian Count: {pedestrian_count}"
        icon_color = 'black'  # Default color
        icon_html = f'<i class="fa fa-user" style="color: {icon_color}; font-size: 24px;"></i>'
        
        # Add the pedestrian marker to the map with an offset
        folium.Marker(
            [lat + pedestrian_offset, lon + pedestrian_offset],
            popup=folium.Popup(popup, max_width=300),
            icon=folium.DivIcon(html=icon_html)
        ).add_to(m)
        

def euclidean_distance(lat1, lon1, lat2, lon2):
    # Radius of the Earth in kilometers
    radius = 6371.0
    
    # Convert latitude and longitude from degrees to radians
    lat1 = math.radians(lat1)
    lon1 = math.radians(lon1)
    lat2 = math.radians(lat2)
    lon2 = math.radians(lon2)
    
    # Haversine formula
    dlon = lon2 - lon1
    dlat = lat2 - lat1
    a = math.sin(dlat / 2)**2 + math.cos(lat1) * math.cos(lat2) * math.sin(dlon / 2)**2
    c = 2 * math.atan2(math.sqrt(a), math.sqrt(1 - a))
    distance = radius * c
    
    return distance

       
 # Add traffic signals to the map
for index, row in traffic_lights.iterrows():
    lat_signal = float(row['latitude'])
    lon_signal = float(row['longitude'])
    site_name = row['SITE_NAME']
    
    # Customize the marker color and icon for traffic signals
    icon_color = 'green'  # Traffic signals are shown in green
    icon_html = f'<i class="fa fa-traffic-light" style="color: {icon_color}; font-size: 24px;"></i>'
    
    # Check if there's a person above 70+ near the specific traffic signal
    if any(ped_age == '70+' and euclidean_distance(lat_signal, lon_signal, ped_lat, ped_lon) < 0.0001 for ped_lat, ped_lon, ped_age in zip(combined_df_cleaned['latitude'], combined_df_cleaned['longitude'], combined_df_cleaned['Age Group'])):
        # Create a pop-up label with the "old man holding a stick" symbol
        popup = f"Traffic Signal: {site_name}<br><i class= 'fa fa-blind' style='color: green; font-size: 24px;'></i><br>Time Extended up to 13 seconds"
    else:
        # Default pop-up label for other traffic signals
        popup = f"Traffic Signal: {site_name}"
    
    # Create a marker with popup for traffic signals (icon remains regular traffic signal)
    folium.Marker(
        [lat_signal, lon_signal],
        popup=folium.Popup(popup, max_width=300),
        icon=folium.DivIcon(html=icon_html)  # Use any default icon, you can change it to 'fa-traffic-light' for a traffic light symbol
    ).add_to(m)

# Display the map
m

Make this Notebook Trusted to load map: File -> Trust Notebook

We have developed an interactive map that displays incidences involving pedestrians who are older  and traffic signals. This map offers important information about driving safety. Locations of traffic signals are shown on the map, and each one is identified by a green traffic light icon. Pedestrian occurrences: Markers on the map indicate where there have been pedestrian occurrences. . Age Group 70+: Red icons that resemble elderly pedestrians who are over the age of 70, indicating greater vulnerability. Age Group 64-69: Blue symbols are used to indicate incidents involving pedestrians aged 64 to 69. Clicking on a marker reveals detailed information about the incident, including the year, location, pedestrian count, and age group. When a traffic light is clicked, a pop-up notification reveals whether there are any pedestrians older than 70 close by. If there are any such pedestrians, a green traffic signal icon and a message stating that the crossing time has been extended by 13 seconds to help them are displayed.

4. Findings and Opportunities

• We have learned important information about pedestrian events, accident hotspots, and road safety issues through a data-driven approach. The strong effect of pedestrian age on accident rates is one important discovery. Accident risk is higher for pedestrians 65 and older. This susceptibility is due to issues including longer reaction times and a decreased capacity to effectively gauge the speed and distance of oncoming cars. We looked at creative answers that were modelled after Singaporean practises to overcome these problems. The implementation of a technology that allows elderly and disabled people to lengthen crossing times by merely pressing a card on the traffic signal is one such approach.

• We can analyse pedestrian incident data in depth to spot trends, taking into account elements like age, location, hour of the day, and weather. This investigation can provide additional light on the factors that contribute to pedestrian accident causes. Involve relevant stakeholders, including traffic safety authorities and advocacy groups for the elderly and disabled, in discussions and planning for road safety improvements. We can also, assess the technical and financial feasibility of implementing solutions like the touch card system for extending crossing times.

5. Conclusion

• In summary, efforts to improve traffic safety through data analysis have produced insightful information and prospects. In-depth analysis of pedestrian events has shown accident hotspots and opportunities for safety enhancement. Our study emphasises the important role that ageing has in affecting accident rates, especially for pedestrians over 65, due to factors like poorer judgement of traffic speed and distance.

• The suggested remedy presents a novel strategy and draws inspiration from international practises like Singapore's. It has the potential to dramatically increase pedestrian safety to allow elderly and disabled walkers to lengthen crossing durations with a simple card tap on traffic signals. We can significantly advance towards a safer and more pedestrian-friendly cityscape by working together and embracing creative solutions, which will eventually save lives and improve the quality of life for those who live in our neighbourhood.

6. Thank You.